Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Mon, 5 Feb 2018 13:30:13 +0800
From:   Aaron Lu <aaron.lu@intel.com>
To:     linux-mm@kvack.org, linux-kernel@vger.kernel.org
Cc:     Andrew Morton <akpm@linux-foundation.org>,
        Huang Ying <ying.huang@intel.com>,
        Dave Hansen <dave.hansen@intel.com>,
        Kemi Wang <kemi.wang@intel.com>,
        Tim Chen <tim.c.chen@linux.intel.com>,
        Andi Kleen <ak@linux.intel.com>,
        Michal Hocko <mhocko@suse.com>,
        Vlastimil Babka <vbabka@suse.cz>,
        Mel Gorman <mgorman@techsingularity.net>,
        Daniel Jordan <daniel.m.jordan@oracle.com>
Subject: RFC: eliminate zone->lock contention for will-it-scale/page_fault1
 on big server
Message-ID: <20180205053013.GB16980@intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20180124023050.20097-1-aaron.lu@intel.com>
User-Agent: Mutt/1.9.1 (2017-09-22)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

In addition the the two patches, there are two more patches that I would
like to get some feedback.

The two patches are more radical: the 3rd deals with free path
zone->lock contention by avoiding doing any merge for order0 pages while
the 4th deals with allocation path zone->lock contention by taking
pcp->batch pages off the free_area order0 list without the need to
iterate the list.

Both patches are developed based on "the most time consuming part of
operations under zone->lock is cache misses on struct page".

The 3rd patch may be controversial but doesn't have correctness problem;
the 4th is in an early stage and serves only as a proof-of-concept.

Your comments are appreciated, thanks.